GLM-5.2
Z.AIOpen weightsReasoningVerifiedReleased Jun 16, 2026Input: text
Overall score
60.5
Rank 55 of 222 · 52.6–68.5
Context window
1M
1,000,000 tokens
Price per 1M tokens
$1.40 in · $4.40 out
Source: BenchLM
Blended price
$2.15
3:1 input:output tokens
Output speed
—
Not measured yet
Category scores
Scores per capability area with the model's rank among ranked models in that category.
- 1Instruction following#2689.9
- 2Reasoning78.8
- 3Math#378.4
- 4Agentic & tool use#3672.1
- 5Knowledge#5465.3
- 6Coding#3540.8
Benchmark results (23)
As reported in system cards, papers and public evaluations. Position compares this model with every other model reporting the same benchmark.
Agentic & tool use
| Terminal-Bench 2.1 | 81.0 | #28 of 68 |
| MCP Atlas | 76.8 | #14 of 42 |
| MazeBench | 0 | #20 of 42 |
| HLE w/ tools | 54.7 | #16 of 27 |
| Toolathlon | 48.2 | #12 of 22 |
| ResearchClawBench | 20.7 | #2 of 19 |
| Terminal-Bench 3.0 | 4.6 | #13 of 13 |
Math
| FrontierMath v2 (Tiers 1-3) | 59.206 | #34 of 64 |
| FrontierMath v2 (Tier 4) | 29.268 | #27 of 51 |
| AIME26 | 99.2 | #1 of 29 |
| HMMT Feb 2026 | 92.5 | #7 of 24 |
| HMMT Nov 2025 | 94.4 | #3 of 12 |
| MMAnswerBench | 91.0 | #1 of 10 |
Coding
| SWE-bench Pro | 62.1 | #22 of 69 |
| NL2Repo | 48.9 | #11 of 31 |
| ProgramBench | 63.7 | #7 of 15 |
| PostTrainBench v1.1 | 31.7 | #10 of 15 |
| OpenHarmony Bench | 58.4 | #5 of 12 |
Knowledge
| GPQA | 91.2 | #19 of 117 |
| HLE | 40.5 | #30 of 69 |
| HLE w/o tools | 40.5 | #22 of 43 |
Multimodal
| Design Arena Website | 1285 | #22 of 91 |
Pricing by source
USD per 1M tokens.
| Source | Input | Output | Cached in |
|---|---|---|---|
| List price (BenchLM) | $1.40 | $4.40 | — |
| OpenRouter z-ai/glm-5.2 | $0.62 | $2.42 | $0.14 |
Z.AI's official pricing page lists GLM-5.2 at $1.40 input / $4.40 output per million tokens.
Arena ratings
Human-preference ratings from LMArena. Best-rated variant shown.
| Text(max) | 1470 | #31 · 49.8K votes |
| WebDev (coding)(Max) | 1604 | #30 · 16.7K votes |
Open weights
From the Hugging Face repository.
- Parameters
- 753B
- Licence
- mit
- Downloads (30 days)
- 391K
- Repository
- zai-org/GLM-5.2 ↗
Speed
Measured by us through OpenRouter (median of trials).
Not measured yet.
Related models
Same family
- GLM-5.156.3
- GLM-554.9
- GLM-5-Turbo53.2
- GLM-5.368.0
Similar score
- GPT-5.3 Codex60.5 · $4.81
- Muse Spark60.5 · —
- GPT-5.259.9 · $4.81
- Gemini 3 Pro61.2 · $4.50
- Hy4 preview61.4 · $1.13